200948647 九、發明說明： 【發明所屬之技術領域】 本發明係關於提高行車安全的方法’尤其是有關於—種利 用影像處理來判斷車身是否發生偏移的方式來提醒駕敬人 的方法。 【先前技術】 ❾ 辅助行車安全、避免駕駛因為疲勞或酒醉而肇事的裝置、 系統、與方法一直疋學界與產業界研發不懈的題目之_。 近年有一個研發方向主要是利用影像處理的技術，辨識出 車道的標線，而在車體偏移出車道而跨越車道標線時，對 駕駛提出提醒或警示的功能，藉此來降低筆事的可能性。 例如，美國發明專利第5,530,420號是採用安裝在後視鏡左 右的二個影像擷取裝置，利用所擷取到的影像以及影像擷 ❹ 取裝置的視角差異來判斷車道標線的位置。美國發明專利 第5,689,249號也是利用能照到整個車道、裝在車内的一個 影像擷取裝置來判斷車道標線。美國發明專利第6,489,887 破則除了用影像掏取褒置以外，還輔以一個角度感測器來 判k/f車子的偏航角度與車道的曲率的關係。此外，我國發 明專利第12280876號則採單一影像擷取裝置，以及特別的 影像處理方法來判斷行車時不當的偏移。其方法需要先設 疋好影像的「搜寻區域_j’然後在一些條件下如果在搜尋區 域找不到標線時發出警示。 5 200948647 這些先前技術有一個共通 位於車身：旁的左右就是都試圖找出車身前方、 生異常時發出警訊。這是:線’然後在二條標線的變化發 的解決方式。這些先前#^種從駕駛人或車㈣觀點出發 雨天照明;f良的情形下、下主要的缺點：⑴在陰 機制時，從掘取的影像中^夜間標線上沒有適當的反射 ❹ 難的；（2)通常需要採用至=確判斷標線的位置，是有困 本較高；以及（3)演算〜個影像錄裝i，因此成 的車用電腦，因此進一步造成^雜^’因此需要較高運算能力 ^顆·兩的成本。 【發明内容】 本發明因此提出一種新穎、有效 方法，來解決先前技術的缺點。 但簡單 而且低成本的 施 的 以 ❹ 本方法首先提供至少一影像擷取装置。在本方法 例裡，影像擷取裝置係裝設於車輛左、右前輪其中:一 前方的車頭上，致使所擷取的影像只包含深色的路面 及集中影像一側之白色或淺色的單一標線。 對於影像擷取裝置擷取的連續影像，本方法計算其中代表 車道標線的白色或淺色像素的水平分佈情形。當發現白色 或淺色像素由影像的一邊往另一邊移動時（亦即白色或淺 色像素的分佈向另一邊增加、或向一邊減少），就代表偏移 6 200948647 的可能發生，而當白色或淺色像素的分佈符合一判斷條件 時，本方法即提供至少聽覺上、對駕駛人的提醒，例如警 報聲、預錄的語音等。 茲配合所附圖示、實施例之詳細說明及申請專利範圍，將 上述及本發明之其他目的與優點詳述於後。然而，當可了 解所附圖示純係為解說本發明之精神而設，不當視為本發 明範疇之定義。有關本發明範疇之定義，請參照所附之申 Ο 請專利範圍。 【實施方式】 本方法係採用影像處理之方式來判斷行車偏移，因此於實 施上通常至少需提供一影像擷取裝置（通常即是一攝影鏡 頭）。在本方法的一實施例裡，如第la圖所示，影像擷取 裝置100係裝設於車輛10右前輪輪胎20的前方的車頭上。 第lb圖所示係該影像擷取裝置所拍攝影像之示意圖。一般 標線1是以白漆繪製於深色的瀝青路面2上，而且一般只 要是適當的照明，二者的對比是相當明顯的。影像擷取裝 置100的安裝位置、拍攝範圍（如虛線所示）應使代表標 線1的淺色區塊集中出現於影像的右半邊。 本方法通常係實施於一運算裝置。該運算裝置可以是車輛 10本身具有的行車電腦、或是另外設置的運算裝置。在某 些實施方式中，影像擷取裝置與運算裝置亦可結合為單一 7 200948647 裝置，一同安裝於車頭上。運算裝置與影像擷取裝置所需 的電力均來自車輛ίο本身的電力系統。 影像擷取裝置100擷取的影像係經由適當的訊號連線交由 運算裝置依據本方法來處理。習知技術判斷行車偏移的方 式，一般是利用影像處理找出代表標線的曲線或線條出 來。這需要相當的運算量。相對之下，本方法判斷行車偏 移的方式就相當簡單，以第lb圖為例，運算裝置只要在連 續影像中發現代表標線的白色或淺色區塊由影像的一邊往 另一邊移動時，就代表可能發生行車的偏移。 本方法之一實施例如第2a〜2c圖所示。如第2a圖所示，本 方法將每一影像水平劃分為相同大小A到F共6區（亦即 每一區具有相同數目的像素），然後計算每一區表示標線的 白色或淺色像素（亦即亮度值在一參考值以上者）在該區 的後度 D—Pwhite/Ptotal， 其中 5 Pwhite 代表該區白色或淺色像 素的總數，而PtQtal代表該區像素的總數。以下，本說明書 以Da、DB、...、DF代表A到F區各區的白色或淺色像素 的密度。 在正常而且穩定的行車狀況下，每一區的影像中表示標線 的白色或淺色像素的密度DA、DB、...、DF基本上會保持不 變（縱有變化也是非常微小或是暫時的）。例如C〜F區白色 或淺色像素的密度Dc、Dd、DE、DF應該非常接近零，而 200948647 da、db則各有相當穩定的數值。當駕駛人打瞌睡或不清醒 而致車子向右偏移時，所擷取到的影像會如第2b圖所示， 不僅Da、DB會增加，原本幾乎為零的Dc開始持續增加， 然後Dd、DE等也會逐一發生同樣的現象。所以從這樣的變 化趨勢就可以判斷是否有行車向右偏移的情形，而所需要 的計算僅在於統計各區的白色或淺色像素的密度。其判斷 條件可以有多種實施方式，例如某一實施例係採當DC2TC (Tc為一臨界值）時，判斷行車發生向右偏移，又例如某 一實施例係採當DC^TC、且行車速度不為零時，判斷行車 發生向右偏移，再例如某一實施例係採當DC^TC、行車速 度不為零、並且此一情況保持一段時間以上時，判斷行車 發生向右偏移。 而當駕駛人打瞌睡或不清醒而致車子向左偏移時，所擷取 到的影像會如第2c圖所示，先是Db會持續減少，然後Da 也會發生同樣的現象。所以從這樣的變化趨勢就可以判斷 是否有行車向左偏移的情形。同樣地，其判斷條件可以有 多種實施方式，例如某一實施例係採當DA£TA、DBSTB( TA、 TB為臨界值）時，判斷行車發生向左偏移，此外也可以結 合行車速度、以及該判斷條件是否持續成立一段時間以上 等，以更穩定或精確的判斷行車發生向左偏移。 本方法之另一實施例如第3 a〜3 c圖所示。如圖所示，本方 法以影像之水平方向為X軸、垂直方向為Y軸，然後計算 200948647 每一水平位置X上，垂直方向的白色或淺色像素的數目 N(x)。第3a圖係依據第2a圖之影像（以虛線表示）所晝 出的N(x)曲線（即圖中較粗黑線者）。當駕駛人打瞌睡或不 清醒而致車子向右或向左偏移時，所擷取到的影像會如第 2b或第2c圖所示，依據本實施例則會得到如第3b或第3c 圖所示的N(x)曲線。對照第3a、3b、3c圖可以看出，當行 車向左或右偏移時，N(x)曲線的折點（亦即垂直方向開始 有白色或淺色像素出現的水平位置xCQrner )也向右移動 (Xcorner增加）或向左移動（Xcorner減少）。所以從這樣的變 化趨勢就可以判斷是否有行車向左偏移的情形。同樣地， 其判斷條件可以有多種實施方式，例如某一實施例係採當 XcornerSTL、或XC〇rne之TR (TL、TR為臨界值）時，判斷行車 發生向右偏移或向左偏移，此外也可以結合行車速度、方 向燈有無啟動、以及該判斷條件是否持續成立一段時間以 上等，以更穩定或精確的判斷行車發生偏移。 不論是採用像素的數目或密度、也不論是否分隔區段，前 述的實施例都是依據影像裡白色或淺色像素在水平方向的 分佈。然後，當此水平分佈符合一判斷條件時，即可認定 為有行車偏移方生。然後此一判斷條件除了依據所擷取的 影像外，還可以搭配其他行車的狀況綜合判斷，例如當時 車速（靜止中的車輛自無判斷的必要、或是車速在顯著減 慢時可能是因為駕駛人有靠邊的意圖）、與偏移方向同一侧 的方向燈有無打開（例如有打開就不認定為偏移，而是認 200948647 定為車輛靠邊或轉向）、符合判斷條件的狀況是否有持續一 段時間、或是連續一定張數的影像都有符合判斷條件（如 果是短暫的發生而又恢復正常的話，就不認定為偏移）等。 接下來，當依據前述方式判斷有偏移發生時，本方法即會 至少聽覺上、對駕駛人的提醒，例如警報聲、預錄的語音 等。這些通常都是透過一警示裝置來達成。警示裝置至少 包含一個喇叭，而該喇叭可以是車輛ίο本身具有的音響系 統的喇叭、或是另外單獨提供。本方法還可以進一步提供 視覺的提醒，例如閃爍的燈號或文字等。這些視覺的警告 可以是顯示於另外提供的燈號、或是顯示於車輛10本身的 儀表板、後照鏡、或是投射於擋風玻璃上等。而當擷取的 影像恢復正常（不符合判斷條件）時，本方法即會自動停 止聽覺或視覺的警示。 請注意到，影像擷取裝置100也可以裝設於車輛10左前輪 輪胎的前方的車頭上，則所擷取的影像會如第lb圖所示， 但係左右顛倒。運算裝置判斷行車偏移的方式也如同第 2a〜2c、3a~3c圖所示，不過也是採左右顛倒的方式進行。 由於極為類似，以下就不贅述。 還有一種可能的實施例是車輛10左、右前輪輪胎的前方的 車頭上，各裝設有一個影像擷取裝置100,然後由本方法同 時但分別處理二個鏡頭所分別擷取的影像，然後依據前述 11 200948647 的方式來判斷行車向左或向右偏移。此一作法成本稍高但 是由於有兩組影像可以進行綜合的研判，誤判的機會可以 顯著降低（例如只有其中之一判斷為向左、或向右偏移時， 本方法不會發出警示）。 第4圖所示係本方法另一實施例之影像擷取裝置之安裝位 置示意圖。如圖所示的車身下方，影像擷取裝置100係裝 設於車身之下，其位置共有四種可能，分別是左、右前輪 輪胎20的前方、或是左、右後輪輪胎20的前方。這四個 位置可以擇一安裝或是全部都安裝。影像擷取裝置100的 安裝位置、拍攝範圍（如虛線所示）應使代表標線1的淺 色區塊集中出現於影像的一邊或一侧。 本方法在提供影像擷取裝置時，可進一步於影像擷取裝置 上、或其附近提供一照明單元1〇1(例如一發光二極體燈）， 以在陰雨天照明不良的情形下、或是夜間標線上沒有適當 的反射機制時提供額外的照明與補光的功能。其投射範圍 基本上與影像擷取裝置100的拍攝範圍一致。 請注意到，本方法的主要精神在於令影像擷取裝置儘量貼 近路面與一側的標線，致使所攝得的影像主要僅包含深色 的路面與單一淺色的標線，而且使標線出現於影像中的一 侧，因此安裝的位置實不以前述為限。由於所擷取的影像 單純，可以大幅減低影像處理的困難度，而且只需要利用 12 200948647 影像裡白色或淺色像素的水平分 置，可 =!有向-方偏移，cr判 以大幅^向行車的安全。 藉由以上較佳具體實施例之詳述，係 =r與精神，而並非以上述所揭二二= 施例來對本發明之範4加以限制。4目反地，其目的 〇 能涵盍各種改變及具相等性的安•本發明所欲 利範圍的範轉内。 辱 【圖式簡單說明】 第la圖所不係本發明一實施例之影像擷取裝置之安裝位置 示意圖。 第lb圖所示係本發明一實施例之影像擷取裝置所拍攝影像 之示意圖。 ❹ 第2a圖所示係本發明一實施例在正常情況下所擷取影像之 示意圖。 第2b圖所示係本發明一實施例在行車向右偏移時所擷取影 像之示意圖。 第2c圖所示係本發明一實施例在行車向左偏移時所操取影 像之示意圖。 第3a圖所示係本發明另一實施例在正常情況下所擷取影像 及運算結果之示意圖。 第3b圖所示係本發明另一實施例在行車向右偏移時所擷取 13 200948647 影像及運算結果之示意圖。 第3c圖所示係本發明另一實施例在行車向左偏移時所擷取 影像及運算結果之示意圖。 第4圖所示係本發明又一實施例之影像擷取裝置之安裝位 置示意圖。 【主要元件符號說明】 1 標線 2 路面 10 車輛 20 輪胎 100 攝影鏡頭 101 照明單元200948647 IX. INSTRUCTIONS OF THE INVENTION: TECHNICAL FIELD OF THE INVENTION The present invention relates to a method for improving driving safety, particularly to a method for using image processing to determine whether a vehicle body is offset or not to alert a person. [Prior Art] 装置 A device, system, and method that assists in driving safety and avoids driving because of fatigue or drunkenness has been the subject of unremitting research in the academic and industrial circles. In recent years, there has been a research and development direction mainly using image processing technology to identify the marking line of the lane, and to remind the driver of the warning when the vehicle body shifts out of the lane and crosses the lane marking. The possibility. For example, U.S. Patent No. 5,530,420 uses two image capturing devices mounted on the left and right of the rear view mirror to determine the position of the lane markings by using the captured image and the difference in viewing angle of the image capturing device. U.S. Patent No. 5,689,249 also utilizes an image capture device that can be photographed throughout the lane and mounted in the vehicle to determine lane markings. In addition to the image capture device, the US invention patent No. 6,489,887 is supplemented by an angle sensor to determine the relationship between the yaw angle of the k/f car and the curvature of the lane. In addition, China Patent No. 12280876 uses a single image capture device and a special image processing method to determine the improper offset during driving. The method needs to first set the "search area _j" of the image and then, under some conditions, issue a warning if the line is not found in the search area. 5 200948647 These prior art have a common location on the body: the left and right sides are trying Find out the warning when the front of the car body is abnormal. This is: the solution of the line 'and then the change of the two marking lines. These previous #^ kinds of rainy day lighting from the driver's or car (four) point of view; The main shortcomings are: (1) In the yin mechanism, there is no proper reflection in the image on the nighttime; (2) It is usually necessary to use the position of = to determine the position of the marking line. High; and (3) calculus ~ video recording i, thus becoming a car computer, thus further causing a miscellaneous ^' therefore requires a higher computing power ^ two · two costs. [The present invention] A novel and effective method to solve the shortcomings of the prior art. However, the simple and low-cost method first provides at least one image capturing device. In the method example, the image capturing device is installed. The system is installed on the left and right front wheels of the vehicle: on the front of the front, so that the captured image only contains the dark road surface and a single white or light colored marking on the side of the concentrated image. The continuous image captured, the method calculates the horizontal distribution of white or light pixels representing the lane markings. When it is found that white or light pixels move from one side of the image to the other (ie, white or light pixels) The distribution increases to the other side, or decreases to one side, which represents the possible occurrence of offset 6 200948647, and when the distribution of white or light pixels meets a judgment condition, the method provides at least an audible, reminder to the driver. For example, an alarm sound, a pre-recorded voice, etc. The above and other objects and advantages of the present invention will be described in detail later in conjunction with the accompanying drawings, the detailed description of the embodiments, and the claims. The illustrations are purely illustrative of the spirit of the invention, and are not considered to be a definition of the scope of the invention. For the definition of the scope of the invention, please refer to the attached application. [Embodiment] The method uses image processing to determine the driving deviation. Therefore, at least one image capturing device (usually a photographic lens) is usually provided for implementation. In an embodiment of the method, As shown in FIG. 1A, the image capturing device 100 is mounted on the front of the front wheel tire 20 of the vehicle 10. Figure lb is a schematic view of the image captured by the image capturing device. The white paint is drawn on the dark asphalt pavement 2, and generally the contrast is quite obvious as long as it is appropriate illumination. The installation position and shooting range of the image capturing device 100 (as indicated by the dotted line) should be representative The light-colored blocks of line 1 are concentrated in the right half of the image. The method is generally implemented in an arithmetic device, which may be a driving computer owned by the vehicle 10 itself or an arithmetic device separately provided. In some embodiments, the image capture device and the computing device can also be combined into a single 7 200948647 device that is mounted on the front of the vehicle. The power required by both the computing device and the image capture device comes from the power system of the vehicle itself. The image captured by the image capture device 100 is processed by the arithmetic device via the appropriate signal connection. The conventional technique for determining the deviation of a vehicle is generally to use image processing to find a curve or line representing a line. This requires a considerable amount of computation. In contrast, the method for determining the deviation of the vehicle is quite simple. Taking the lb diagram as an example, the computing device only needs to find that the white or light color block representing the reticle moves from one side of the image to the other side in the continuous image. , it represents the deviation of the possible driving. One embodiment of the method is shown in Figures 2a to 2c. As shown in Fig. 2a, the method divides each image level into 6 regions of the same size A to F (i.e., each region has the same number of pixels), and then calculates each region to represent the white or light color of the marked line. The pixel (i.e., the luminance value above a reference value) is the latter degree D-Pwhite/Ptotal of the region, where 5 Pwhite represents the total number of white or light pixels in the region, and PtQtal represents the total number of pixels in the region. Hereinafter, in the present specification, Da, DB, ..., DF represents the density of white or light-colored pixels in each of the areas A to F. Under normal and stable driving conditions, the density DA, DB, ..., DF of the white or light-colored pixels indicating the markings in the image of each zone will remain basically the same (the change is also very small or temporary). For example, the density of D or D, Dd, DE, and DF of white or light pixels in the C to F area should be very close to zero, and 200948647 da and db each have a fairly stable value. When the driver is dozing or not awake and the car is shifted to the right, the captured image will be as shown in Figure 2b. Not only will Da and DB increase, but the almost zero Dc will continue to increase, then Dd. , DE, etc. will also happen the same phenomenon one by one. Therefore, from such a change trend, it can be judged whether or not there is a situation in which the driving is shifted to the right, and the calculation required is only to count the density of white or light pixels in each area. The determination condition may be implemented in various embodiments. For example, when a DC2TC (Tc is a critical value) is used, it is determined that the driving occurs to the right, and for example, a certain embodiment is used as DC^TC, and the driving is performed. When the speed is not zero, it is judged that the driving is shifted to the right, and for example, in a certain embodiment, DC^TC is adopted, the driving speed is not zero, and when the situation is maintained for more than a period of time, it is judged that the driving is shifted to the right. . When the driver is dozing or not awake and the car is shifted to the left, the captured image will be as shown in Figure 2c. First, Db will continue to decrease, and then Da will also have the same phenomenon. Therefore, from such a trend, it can be judged whether there is a situation in which the vehicle is shifted to the left. Similarly, the judgment condition may have various implementation manners. For example, when a certain embodiment adopts DA£TA, DBSTB (TA, TB is a critical value), it is determined that the driving is shifted to the left, and the driving speed may be combined. And whether the judgment condition is continuously established for a period of time or the like, and the vehicle shifts to the left in a more stable or accurate judgment. Another embodiment of the method is shown in Figures 3a to 3c. As shown in the figure, the method uses the horizontal direction of the image as the X-axis and the vertical direction as the Y-axis, and then calculates the number N(x) of white or light-colored pixels in the vertical direction at each horizontal position X of 200948647. Figure 3a is an N(x) curve (i.e., a thicker black line in the figure) drawn from the image of Figure 2a (shown in phantom). When the driver is dozing or not awake and the car is shifted to the right or left, the captured image will be as shown in Figure 2b or 2c. According to this embodiment, the 3b or 3c will be obtained. The N(x) curve shown in the figure. As can be seen from Fig. 3a, 3b, and 3c, when the driving is shifted to the left or right, the vertices of the N(x) curve (that is, the horizontal position xCQrner where white or light pixels appear in the vertical direction) are also Move right (Xcorner increases) or move left (Xcorner decreases). Therefore, it is possible to judge whether or not there is a situation in which the vehicle is shifted to the left from such a change trend. Similarly, the judgment condition may have various implementation manners. For example, when an embodiment adopts XcornerSTL or XC〇rne TR (TL, TR is a critical value), it is determined that the driving occurs to shift to the right or to the left. In addition, it is also possible to combine the driving speed, the presence or absence of the directional light, and whether the judgment condition is continued for a certain period of time or the like, so as to more accurately or accurately determine the deviation of the driving. Regardless of whether the number or density of pixels, or whether the segments are separated, the foregoing embodiments are based on the horizontal distribution of white or light pixels in the image. Then, when the horizontal distribution meets a judgment condition, it can be considered as having a driving deviation. Then, in addition to the captured image, this judgment condition can be combined with the judgment of other driving conditions, such as the speed of the vehicle at that time (the vehicle in the stationary state is necessary for no judgment, or the vehicle speed may be significantly slowed down because of driving) The person has the intention of the edge), whether the direction light on the same side as the offset direction is open (for example, if it is opened, it is not recognized as an offset, but the 200948647 is determined to be the side or the steering of the vehicle), and the condition that meets the judgment condition is continued for a while. , or a certain number of consecutive images have met the judgment conditions (if it is a short-term occurrence and return to normal, it is not recognized as an offset). Next, when it is judged that the offset occurs according to the foregoing manner, the method at least audibly alerts the driver, such as an alarm sound, a pre-recorded voice, and the like. These are usually achieved through a warning device. The warning device includes at least one speaker, and the speaker may be a speaker of the sound system of the vehicle ίο, or separately provided. The method can further provide visual reminders, such as flashing lights or text. These visual warnings may be displayed on a separately provided light number, or displayed on the dashboard of the vehicle 10 itself, a rear view mirror, or projected onto a windshield. When the captured image returns to normal (not meeting the judgment condition), the method automatically stops the hearing or visual warning. Please note that the image capturing device 100 can also be mounted on the front of the front tire of the vehicle 10, and the captured image will be as shown in Figure lb, but inverted. The manner in which the arithmetic unit determines the deviation of the vehicle is also as shown in Figs. 2a to 2c and 3a to 3c, but it is also performed in such a manner that the left and right are reversed. Because it is very similar, the following will not go into details. Another possible embodiment is that the image capturing device 100 is mounted on the front of the front left and right front tires of the vehicle 10, and then the image captured by the two lenses is processed simultaneously by the method, and then respectively According to the manner of the aforementioned 11 200948647, it is judged that the driving is shifted to the left or to the right. This method is slightly more expensive, but because there are two sets of images that can be comprehensively judged, the chance of misjudgment can be significantly reduced (for example, this method does not issue a warning when only one of them judges to be left or right). Figure 4 is a schematic view showing the mounting position of the image capturing device of another embodiment of the method. Under the vehicle body as shown, the image capturing device 100 is mounted under the vehicle body, and its position has four possibilities, namely, the front of the left and right front tires 20, or the front of the left and right rear tires 20, respectively. . These four locations can be installed either individually or all. The mounting position and shooting range of the image capturing device 100 (as indicated by a broken line) should be such that the light blocks representing the marking 1 are concentrated on one side or one side of the image. When the image capturing device is provided, the illumination unit 1〇1 (for example, a light-emitting diode lamp) may be further provided on or near the image capturing device to be in poor lighting conditions during rainy days, or Provides additional illumination and fill light when there is no proper reflection mechanism on the night line. The projection range is substantially the same as the shooting range of the image capturing device 100. Please note that the main spirit of the method is to make the image capturing device as close as possible to the marking on the road surface and on one side, so that the captured image mainly contains only the dark road surface and the single light color marking line, and the marking line is made. Appears on one side of the image, so the location of the installation is not limited to the above. Since the captured image is simple, the difficulty of image processing can be greatly reduced, and only the horizontal division of white or light pixels in the 12 200948647 image can be used, and the directional-square offset can be corrected. To the safety of driving. The specification of the present invention is limited by the following detailed description of the preferred embodiments, and is not intended to limit the scope of the invention. The purpose of 4 反 , , , , , , , , 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic view showing the installation position of an image capturing device according to an embodiment of the present invention. Figure lb is a schematic diagram of an image captured by an image capture device in accordance with an embodiment of the present invention. ❹ Figure 2a is a schematic diagram of an image taken under normal conditions in an embodiment of the present invention. Fig. 2b is a schematic view showing an image taken when the vehicle is shifted to the right in an embodiment of the present invention. Fig. 2c is a schematic view showing an image taken when the vehicle is shifted to the left in an embodiment of the present invention. Fig. 3a is a schematic diagram showing images and operation results taken under normal conditions in another embodiment of the present invention. FIG. 3b is a schematic diagram of the image and operation result of 13 200948647 taken when the vehicle is shifted to the right according to another embodiment of the present invention. Fig. 3c is a schematic view showing an image taken and an operation result when the vehicle is shifted to the left in another embodiment of the present invention. Fig. 4 is a view showing the mounting position of an image capturing device according to still another embodiment of the present invention. [Main component symbol description] 1 Marking line 2 Pavement 10 Vehicle 20 Tire 100 Photographic lens 101 Lighting unit